ProbeX/Model-J__MAE__model_idx_0005

Model-J: MAE Model (model_idx_0005)

This model is part of the Model-J dataset, introduced in:

Learning on Model Weights using Tree Experts (CVPR 2025) by Eliahu Horwitz*, Bar Cavia*, Jonathan Kahana*, Yedid Hoshen

🌐 Project | 📃 Paper | 💻 GitHub | 🤗 Dataset

Model Details

Attribute	Value
Subset	MAE
Split	val
Base Model	facebook/vit-mae-base
Dataset	CIFAR100 (50 classes)

Training Hyperparameters

Parameter	Value
Learning Rate	3e-05
LR Scheduler	constant_with_warmup
Epochs	6
Max Train Steps	1998
Batch Size	64
Weight Decay	0.009
Seed	5
Random Crop	True
Random Flip	False

Performance

Metric	Value
Train Accuracy	0.9743
Val Accuracy	0.8525
Test Accuracy	0.8520

Training Categories

The model was fine-tuned on the following 50 CIFAR100 classes:

beetle, lobster, clock, chair, sea, cloud, road, seal, worm, whale, lawn_mower, boy, otter, television, shrew, crocodile, forest, bus, snail, kangaroo, rocket, can, table, mouse, leopard, rabbit, cup, crab, wolf, squirrel, trout, bicycle, elephant, willow_tree, ray, turtle, snake, aquarium_fish, telephone, dolphin, porcupine, tiger, castle, lamp, raccoon, camel, palm_tree, fox, beaver, couch